Arrant Pedantry


Language, Logic, and Correctness

In “Why Descriptivists Are Usage Liberals”, I said that there some logical problems with declaring something to be right or wrong based on evidence. A while back I explored this problem in a piece titled “What Makes It Right?” over on Visual Thesaurus.

The terms prescriptive and descriptive were borrowed from philosophy, where they are used to talk about ethics, and the tension between these two approaches is reflected in language debates today. The questions we have today about correct usage are essentially the same questions philosophers have been debating since the days of Socrates and Plato: what is right, and how do we know?

As I said on Visual Thesaurus, all attempts to answer these questions run into a fundamental logical problem: just because something is doesn’t mean it ought to be. Most people are uncomfortable with the idea of moral relativism and believe at some level that there must be some kind of objective truth. Unfortunately, it’s not entirely clear just where we find this truth or how objective it really is, but we at least operate under the convenient assumption that it exists.

But things get even murkier when we try to apply this same assumption to language. While we may feel safe saying that murder is wrong and would still be wrong even if a significant portion of the population committed murder, we can’t safely make similar arguments about language. Consider the word bird. In Old English, the form of English spoken from about 500 AD to about 1100 AD, the word was brid. Bird began as a dialectal variant that spread and eventually supplanted brid as the standard form by about 1600. Have we all been saying this word wrong for the last four hundred years or so? Is saying bird just as wrong as saying nuclear as nucular?

No, of course not. Even if it had been considered an error once upon a time, it’s not an error anymore. Its widespread use in Standard English has made it standard, while brid would now be considered an error (if someone were to actually use it). There is no objectively correct form of the word that exists independent of its use. That is, there is no platonic form of the language, no linguistic Good to which a grammarian-king can look for guidance in guarding the city.

This is why linguistics is at its core an empirical endeavor. Linguists concern themselves with investigating linguistic facts, not with making value judgements about what should be considered correct or incorrect. As I’ve said before, there are no first principles from which we can determine what’s right and wrong. Take, for example, the argument that you should use the nominative form of pronouns after a copula verb. Thus you should say It is I rather than It is me. But this argument assumes as prior the premise that copula verbs work this way and then deduces that anything that doesn’t work this way is wrong. Where would such a putative rule come from, and how do we know it’s valid?

Linguists often try to highlight the problems with such assumptions by pointing out, for example, that French requires an object pronoun after the copula (in French you say c’est moi [it’s me], not c’est je [it’s I]) or that English speakers, including renowned writers, have long used object forms in this position. That is, there is no reason to suppose that this rule has to exist, because there are clear counterexamples. But then, as I said before, some linguists leave the realm of strict logic and argue that if everyone says it’s me, then it must be correct.

Some people then counter by calling this argument fallacious, and strictly speaking, it is. Mededitor has called this the Jane Austen fallacy (if Jane Austen or some other notable past writer has done it, then it must be okay), and one commenter named Kevin S. has made similar arguments in the comments on Kory Stamper’s blog, Harmless Drudgery.

There, Kevin S. attacked Ms. Stamper for noting that using lay in place of lie dates at least to the days of Chaucer, that it is very common, and that it “hasn’t managed to destroy civilization yet.” These are all objective facts, yet Kevin S. must have assumed that Ms. Stamper was arguing that if it’s old and common, it must be correct. In fact, she acknowledged that it is nonstandard and didn’t try to argue that it wasn’t or shouldn’t be. But Kevin S. pointed out a few fallacies in the argument that he assumed that Ms. Stamper was making: an appeal to authority (if Chaucer did it, it must be okay), the “OED fallacy” (if it has been used that way in the past, it must be correct), and the naturalistic fallacy, which is deriving an ought from an is (lay for lie is common; therefore it ought to be acceptable).

And as much as I hate to say it, technically, Kevin S. is right. Even though he was responding to an argument that hadn’t been made, linguists and lexicographers do frequently make such arguments, and they are in fact fallacies. (I’m sure I’ve made such arguments myself.) Technically, any argument that something should be considered correct or incorrect isn’t a logical argument but a persuasive one. Again, this goes back to the basic difference between descriptivism and prescriptivism. We can make statements about the way English appears to work, but making statements about the way English should work or the way we think people should feel about it is another matter.

It’s not really clear what Kevin S.’s point was, though, because he seemed to be most bothered by Ms. Stamper’s supposed support of some sort of flabby linguistic relativism. But his own implied argument collapses in a heap of fallacies itself. Just as we can’t necessarily call something correct just because it occurred in history or because it’s widespread, we can’t necessarily call something incorrect just because someone invented a rule saying so.

I could invent a rule saying that you shouldn’t ever use the word sofa because we already have the perfectly good word couch, but you would probably roll your eyes and say that’s stupid because there’s nothing wrong with the word sofa. Yet we give heed to a whole bunch of similarly arbitrary rules invented two or three hundred years ago. Why? Technically, they’re no more valid or logically sound than my rule.

So if there really is such a thing as correctness in language, and if any argument about what should be considered correct or incorrect is technically a logical fallacy, then how can we arrive at any sort of understanding of, let alone agreement on, what’s correct?

This fundamental inability to argue logically about language is a serious problem, and it’s one that nobody has managed to solve or, in my opinion, ever will completely solve. This is why the war of the scriptivists rages on with no end in sight. We see the logical fallacies in our opponents’ arguments and the flawed assumptions underlying them, but we don’t acknowledge—or sometimes even see—the problems with our own. Even if we did, what could we do about them?

My best attempt at an answer is that both sides simply have to learn from each other. Language is a democracy, true, but, just like the American government, it is not a pure democracy. Some people—including editors, writers, English teachers, and usage commentators—have a disproportionate amount of influence. Their opinions carry more weight because people care what they think.

This may be inherently elitist, but it is not necessarily a bad thing. We naturally trust the opinions of those who know the most about a subject. If your car won’t start, you take it to a mechanic. If your tooth hurts, you go to the dentist. If your writing has problems, you ask an editor.

Granted, using lay for lie is not bad in the same sense that a dead starter motor or an abscessed tooth is bad: it’s a problem only in the sense that some judge it to be wrong. Using lay for lie is perfectly comprehensible, and it doesn’t violate some basic rule of English grammar such as word order. Furthermore, it won’t destroy the language. Just as we have pairs like lay and lie or sit and set, we used to have two words for hang, but nobody claims that we’ve lost a valuable distinction here by having one word for both transitive and intransitive uses.

Prescriptivists want you to know that people will judge you for your words (and—let’s be honest—usually they’re the ones doing the judging), and descriptivists want you to soften those judgements or even negate them by injecting them with a healthy dose of facts. That is, there are two potential fixes for the problem of using words or constructions that will cause people to judge you: stop using that word or construction, or get people to stop judging you and others for that use.

In reality, we all use both approaches, and, more importantly, we need both approaches. Even most dyed-in-the-wool prescriptivists will tell you that the rule banning split infinitives is bogus, and even most liberal descriptivists will acknowledge that if you want to be taken seriously, you need to use Standard English and avoid major errors. Problems occur when you take a completely one-sided approach, insisting either that something is an error even if almost everyone does it or that something isn’t an error even though almost everyone rejects it. In other words, good usage advice has to consider not only the facts of usage but speakers’ opinions about usage.

For instance, you can recognize that irregardless is a word, and you can even argue that there’s nothing technically wrong with it because nobody cares that the verbs bone and debone mean the same thing, but it would be irresponsible not to mention that the word is widely considered an error in educated speech and writing. Remember that words and constructions are not inherently correct or incorrect and that mere use does not necessarily make something correct; correctness is a judgement made by speakers of the language. This means that, paradoxically, something can be in widespread use even among educated speakers and can still be considered an error.

This also means that on some disputed items, there may never be anything approaching consensus. While the facts of usage may be indisputable, opinions may still be divided. Thus it’s not always easy or even possible to label something as simply correct or incorrect. Even if language is a democracy, there is no simple majority rule, no up and down vote to determine whether something is correct. Something may be only marginally acceptable or correct only in certain situations or according to certain people.

But as in a democracy, it is important for people to be informed before metaphorically casting their vote. Bryan Garner argues in his Modern American Usage that what people want in language advice is authority, and he’s certainly willing to give it to you. But I think what people really need is information. For example, you can state authoritatively that regardless of past or present usage, singular they is a grammatical error and always will be, but this is really an argument, not a statement of fact. And like all arguments, it should be supported with evidence. An argument based solely or primarily on one author’s opinion—or even on many people’s opinions—will always be a weaker argument than one that considers both facts and opinion.

This doesn’t mean that you have to accept every usage that’s supported by evidence, nor does it mean that all evidence is created equal. We’re all human, we all still have opinions, and sometimes those opinions are in defiance of facts. For example, between you and I may be common even in educated speech, but I will probably never accept it, let alone like it. But I should not pretend that my opinion is fact, that my arguments are logically foolproof, or that I have any special authority to declare it wrong. I think the linguist Thomas Pyles said it best:

Too many of us . . . would seem to believe in an ideal English language, God-given instead of shaped and molded by man, somewhere off in a sort of linguistic stratosphere—a language which nobody actually speaks or writes but toward whose ineffable standards all should aspire. Some of us, however, have in our worst moments suspected that writers of handbooks of so-called “standard English usage” really know no more about what the English language ought to be than those who use it effectively and sometimes beautifully. In truth, I long ago arrived at such a conclusion: frankly, I do not believe that anyone knows what the language ought to be. What most of the authors of handbooks do know is what they want English to be, which does not interest me in the least except as an indication of the love of some professors for absolute and final authority.[1]

In usage, as in so many other things, you have to learn to live with uncertainty.

  1. [1] “Linguistics and Pedagogy: The Need for Conciliation,” in Selected Essays on English Usage, ed. John Algeo (Gainesville: University Presses of Florida, 1979), 169–70.


Why Descriptivists Are Usage Liberals

Outside of linguistics, the people who care most about language tend to be prescriptivists—editors, writers, English teachers, and so on—while linguists and lexicographers are descriptivists. “Descriptive, not prescriptive!” is practically the linguist rallying cry. But we linguists have done a terrible job of explaining just what that means and why it matters. As I tried to explain in “What Descriptivism Is and Isn’t”, descriptivism is essentially just an interest in facts. That is, we make observations about what the language is rather than state opinions about how we’d like it to be.

Descriptivism is often cast as the opposite of prescriptivism, but they aren’t opposites at all. But no matter how many times we insist that “descriptivism isn’t ‘anything goes’”, people continue to believe that we’re all grammatical anarchists and linguistic relativists, declaring everything correct and saying that there’s no such thing as a grammatical error.

Part of the problem is that whenever you conceive of two approaches as opposing points of view, people will assume that they’re opposite in every regard. Prescriptivists generally believe that communication is important, that having a standard form of the language facilitates communication, and that we need to uphold the rules to maintain the standard. And what people often see is that linguists continually tear down the rules and say that they don’t really matter. The natural conclusion for many people is that linguists don’t care about maintaining the standard or supporting good communication—they want a linguistic free-for-all instead. Then descriptivists appear to be hypocrites for using the very standard they allegedly despise.

It’s true that many descriptivists oppose rules that they disagree with, but as I’ve said before, this isn’t really descriptivism—it’s anti-prescriptivism, for lack of a better term. (Not because it’s the opposite of prescriptivism, but because it often prescribes the opposite of what traditional linguistic prescriptivism does.) Just ask yourself how an anti-prescriptive sentiment like “There’s nothing wrong with singular they” is a description of linguistic fact.

So if that’s not descriptivism, then why do so many linguists have such liberal views on usage? What does being against traditional rules have to do with studying language? And how can linguists oppose rules and still be in favor of good communication and Standard English?

The answer, in a nutshell, is that we don’t think that the traditional rules have much to do with either good communication or Standard English. The reason why we think that is a little more complicated.

Linguists have had a hard time defining just what Standard English is, but there are several ideas that recur in attempts to define it. First, although Standard English can certainly be spoken, it is often conceived of as a written variety, especially in the minds of non-linguists. Second, it is generally more formal, making it appropriate for a wide range of serious topics. Third, it is educated, or rather, it is used by educated speakers. Fourth, it is supraregional, meaning that it is not tied to a specific region, as most dialects are, but that it can be used across an entire language area. And fifth, it is careful or edited. Notions of uniformity and prestige are often thrown into the mix as well.

Careful is a vague term, but it means that users of Standard English put some care into what they say or write. This is especially true of most published writing; the entire profession of editing is dedicated to putting care into the written word. So it’s tempting to say that following the rules is an important part of Standard English and that tearing down those rules tears down at least that part of Standard English.

But the more important point is that Standard English is ultimately rooted in the usage of actual speakers and writers. It’s not just that there no legislative body declaring what’s standard, but that there are no first principles from which we can deduce what’s standard. All languages are different, and they change over time, so how can we know what’s right or wrong except by looking at the evidence? This is what descriptivists try to do when discussing usage: look at the evidence from historical and current usage and draw meaningful conclusions about what’s right or wrong. (There are some logical problems with this, but I’ll address those another time.)

Let’s take singular they, for example. The evidence shows that it’s been in use for centuries not just by common folk or educated speakers but by well-respected writers from Geoffrey Chaucer to Jane Austen. The evidence also shows that it’s used in fairly predictable ways, generally to refer to indefinite pronouns or to nouns that don’t specify gender. Its use has not caused the grammar of English to collapse, and it seems like a rather felicitous solution to the gender-neutral pronoun problem. So at least from a dispassionate linguistic point of view, there is no problem with it.

From another point of view, though, there is something wrong with it: some people don’t like it. This is a social rather than a linguistic fact, but it’s a fact nonetheless. But this social fact arose because at some point someone declared—contrary to the linguistic facts—that singular they is a grammatical error that should be avoided. Here’s where descriptivists depart from description and get into anti-prescription. If people have been taught to dislike this usage, it stands to reason that they could be taught to get over this dislike.

That is, linguists are engaging in anti-prescriptivism to counter the prescriptivism that isn’t rooted in linguistic fact. So when they debunk or tear down traditional rules, it’s not that they don’t value Standard English or good communication; it’s that they think that those particular rules have nothing to do with either.

To be fair, I think that many linguists think they’re still merely describing when they’re countering prescriptive attitudes. Saying that singular they has been used for centuries by respected writers, that it appears to follow fairly well-defined rules, and that the proscription against it is not based in linguistic fact is descriptive; saying that people need to get over their dislike and accept it is not.

And this is precisely why I think descriptivism and prescriptivism not only can but should coexist. It’s not wrong to have opinions on what’s right or wrong, but I think it’s better if those opinions have some basis in fact. Guidance on issues of usage can really only be relevant and valid if it takes all the evidence into account—who uses a certain word of construction, in what circumstances, and so on. These are all facts that can be investigated, and linguistics provides a solid methodological framework for doing so. Anything that ignores the facts reduces to one sort of ipse dixit or another, either a statement from an authority declaring something to be right or wrong or one’s own preferences or pet peeves.

Linguists value good communication, and we recognize the importance of Standard English. But our opinions on both are informed by our study of language and by our emphasis on facts and evidence. This isn’t “anything goes”, or at least no more so than language has always been. People have always worried about language change, but language has always turned out fine. Inventing new rules to try to regulate language will not save it from destruction, and tossing out the rules that have no basis in fact will not hasten the language’s demise. But recognizing that some rules don’t matter may alleviate some of those worries, and I think that’s a good thing for both camps.


Lynne Truss and Chicken Little

Lynne Truss, author of the bestselling Eats, Shoots & Leaves: The Zero Tolerance Approach to Punctuation, is at it again, crying with her characteristic hyperbole and lack of perspective that the linguistic sky is falling because she got a minor bump on the head.

As usual, Truss hides behind the it’s-just-a-joke-but-no-seriously defense. She starts by claiming to have “an especially trivial linguistic point to make” but then claims that the English language is doomed, and it’s all linguists’ fault. According to Truss, linguists have sat back and watched while literacy levels have declined—and have profited from doing so.

What exactly is the problem this time? That some people mistakenly write some phrases as compound words when they’re not, such as maybe for may be or anyday for any day. (This isn’t even entirely true; anyday is almost nonexistent in print, even in American English, according to Google Ngram Viewer.) I guess from anyday it’s a short, slippery slope to complete language chaos, and then “we might as well all go off and kill ourselves.”

But it’s not clear what her complaint about erroneous compound words has to do with literacy levels. If the only problem with literacy is that some people write maybe when they mean may be, then it seems to be, as she originally says, an especially trivial point. Yes, some people deviate from standard orthography. While this may be irritating and may occasionally cause confusion, it’s not really an indication that people don’t know how to read or write. Even educated people make mistakes, and this has always been the case. It’s not a sign of impending doom.

But let’s consider the analogies she chose to illustrate linguists’ supposed negligence. She says that we’re like epidemiologists who simply catalog all the ways in which people die from diseases or like architects who make notes while buildings collapse. (Interestingly, she makes two remarks about how well paid linguists are. Of course, professors don’t actually make that much, especially those in the humanities or social sciences. And it smacks of hypocrisy from someone whose book has sold 3 million copies.)

Perhaps there is a minor crisis in literacy, at least in the UK. This article says that 16–24-year-olds in the UK are lagging behind many counterparts in other first-world countries. (The headline suggests that they’re trailing the entire world, but the study only looked at select countries from Europe and east Asia.) Wikipedia, however, says that the UK has a 99 percent literacy rate. Maybe young people are slipping a bit, and this is certainly something that educators should address, but it doesn’t appear that countless people are dying from an epidemic of slightly declining literacy rates or that our linguistic structures are collapsing. This is simply not the linguistic apocalypse that Truss makes it out to be.

Anyway, even if it were, why would it be linguists’ job to do something about it? Literacy is taught in primary and secondary school and is usually the responsibility of reading, language arts, or English teachers—not linguists. Why not criticize English professors for sitting back and collecting fat paychecks for writing about literary theory while our kids struggle to read? Because they’re not her ideological enemy, that’s why. Linguists often oppose language pedants like Truss, and so Truss finds some reason—contrived though it may be—to blame them. Though some applied linguists do in fact study things like language acquisition and literacy, most linguists hew to the more abstract and theoretical side of language—syntax, morphology, phonology, and so on. Blaming descriptive linguists for children’s illiteracy is like blaming physicists for children’s inability to ride bikes.

And maybe the real reason why linguists are unconcerned about the upcoming linguistic apocalypse is that there simply isn’t one. Maybe linguists are like meteorologists who observe that, contrary to the claims of some individuals, the sky is not actually falling. In studying the structure of other languages and the ways in which languages change, linguists have realized that language change is not decay. Consider the opening lines from Beowulf, an Old English epic poem over a thousand years old:

HWÆT, WE GAR-DEna in geardagum,
þeodcyninga þrym gefrunon,
hu ða æþelingas ellen fremedon!

Only two words are instantly recognizable to modern English speakers: we and in. The changes from Old English to modern English haven’t made the language better or worse—just different. Some people maintain that they understand that language changes but say that they still oppose certain changes that seem to come from ignorance or laziness. They fear that if we’re not vigilant in opposing such changes, we’ll lose our ability to communicate. But the truth is that most of those changes from Old English to modern English also came from ignorance or laziness, and we seem to communicate just fine today.

Languages can change very radically over time, but contrary to popular belief, they never devolve into caveman grunting. This is because we all have an interest in both understanding and being understood, and we’re flexible enough to adapt to changes that happen within our lifetime. And with language, as opposed to morality or ethics, there is no inherent right or wrong. Correct language is, in a nutshell, what its users consider to be correct for a given time, place, and audience. One generation’s ignorant change is sometimes the next generation’s proper grammar.

It’s no surprise that Truss fundamentally misunderstands what linguists and lexicographers do. She even admits that she was “seriously unqualified” for linguistic debate a few years back, and it seems that nothing has changed. But that probably won’t stop her from continuing to prophesy the imminent destruction of the English language. Maybe Truss is less like Chicken Little and more like the boy who cried wolf, proclaiming disaster not because she actually sees one coming, but rather because she likes the attention.


My Thesis

I’ve been putting this post off for a while for a couple of reasons: first, I was a little burned out and was enjoying not thinking about my thesis for a while, and second, I wasn’t sure how to tackle this post. My thesis is about eighty pages long all told, and I wasn’t sure how to reduce it to a manageable length. But enough procrastinating.

The basic idea of my thesis was to see which usage changes editors are enforcing in print and thus infer what kind of role they’re playing in standardizing (specifically codifying) usage in Standard Written English. Standard English is apparently pretty difficult to define precisely, but most discussions of it say that it’s the language of educated speakers and writers, that it’s more formal, and that it achieves greater uniformity by limiting or regulating the variation found in regional dialects. Very few writers, however, consider the role that copy editors play in defining and enforcing Standard English, and what I could find was mostly speculative or anecdotal. That’s the gap my research aimed to fill, and my hunch was that editors were not merely policing errors but were actively introducing changes to Standard English that set it apart from other forms of the language.

Some of you may remember that I solicited help with my research a couple of years ago. I had collected about two dozen manuscripts edited by student interns and then reviewed by professionals, and I wanted to increase and improve my sample size. Between the intern and volunteer edits, I had about 220,000 words of copy-edited text. Tabulating the grammar and usage changes took a very long time, and the results weren’t as impressive as I’d hoped they’d be. There were still some clear patterns, though, and I believe they confirmed my basic idea.

The most popular usage changes were standardizing the genitive form of names ending in -s (Jones’>Jones’s), which>that, towards>toward, moving only, and increasing parallelism. These changes were not only numerically the most popular, but they were edited at fairly high rates—up to 80 percent. That is, if towards appeared ten times, it was changed to toward eight times. The interesting thing about most of these is that they’re relatively recent inventions of usage writers. I’ve already written about which hunting on this blog, and I recently wrote about towards for Visual Thesaurus.

In both cases, the rule was invented not to halt language change, but to reduce variation. For example, in unedited writing, English speakers use towards and toward with roughly equal frequency; in edited writing, toward outnumbers towards 10 to 1. With editors enforcing the rule in writing, the rule quickly becomes circular—you should use toward because it’s the norm in Standard (American) English. Garner used a similarly circular defense of the that/which rule in this New York Times Room for Debate piece with Robert Lane Greene:

But my basic point stands: In American English from circa 1930 on, “that” has been overwhelmingly restrictive and “which” overwhelmingly nonrestrictive. Strunk, White and other guidebook writers have good reasons for their recommendation to keep them distinct — and the actual practice of edited American English bears this out.

He’s certainly correct in saying that since 1930 or so, editors have been changing restrictive which to that. But this isn’t evidence that there’s a good reason for the recommendation; it’s only evidence that editors believe there’s a good reason.

What is interesting is that usage writers frequently invoke Standard English in defense of the rules, saying that you should change towards to toward or which to that because the proscribed forms aren’t acceptable in Standard English. But if Standard English is the formal, nonregional language of educated speakers and writers, then how can we say that towards or restrictive which are nonstandard? What I realized is this: part of the problem with defining Standard English is that we’re talking about two similar but distinct things—the usage of educated speakers, and the edited usage of those speakers. But because of the very nature of copy editing, we conflate the two. Editing is supposed to be invisible, so we don’t know whether what we’re seeing is the author’s or the editor’s.

Arguments about proper usage become confused because the two sides are talking past each other using the same term. Usage writers, editors, and others see linguists as the enemies of Standard (Edited) English because they see them tearing down the rules that define it, setting it apart from educated but unedited usage, like that/which and toward/towards. Linguists, on the other hand, see these invented rules as being unnecessarily imposed on people who already use Standard English, and they question the motives of those who create and enforce the rules. In essence, Standard English arises from the usage of educated speakers and writers, while Standard Edited English adds many more regulative rules from the prescriptive tradition.

My findings have some serious implications for the use of corpora to study usage. Corpus linguistics has done much to clarify questions of what’s standard, but the results can still be misleading. With corpora, we can separate many usage myths and superstitions from actual edited usage, but we can’t separate edited usage from simple educated usage. We look at corpora of edited writing and think that we’re researching Standard English, but we’re unwittingly researching Standard Edited English.

None of this is to say that all editing is pointless, or that all usage rules are unnecessary inventions, or that there’s no such thing as error because educated speakers don’t make mistakes. But I think it’s important to differentiate between true mistakes and forms that have simply been proscribed by grammarians and editors. I don’t believe that towards and restrictive which can rightly be called errors, and I think it’s even a stretch to call them stylistically bad. I’m open to the possibility that it’s okay or even desirable to engineer some language changes, but I’m unconvinced that either of the rules proscribing these is necessary, especially when the arguments for them are so circular. At the very least, rules like this serve to signal to readers that they are reading Standard Edited English. They are a mark of attention to detail, even if the details in question are irrelevant. The fact that someone paid attention to them is perhaps what is most important.

And now, if you haven’t had enough, you can go ahead and read the whole thesis here.


Relative Pronoun Redux

A couple of weeks ago, Geoff Pullum wrote on Lingua Franca about the that/which rule, which he calls “a rule which will live in infamy”. (For my own previous posts on the subject, see here, here, and here.) He runs through the whole gamut of objections to the rule—that the rule is an invention, that it started as a suggestion and became canonized as grammatical law, that it has “an ugly clutch of exceptions”, that great writers (including E. B. White himself) have long used restrictive which, and that it’s really the commas that distinguish between restrictive and nonrestrictive clauses, as they do with other relative pronouns like who.

It’s a pretty thorough deconstruction of the rule, but in a subsequent Language Log post, he despairs of converting anyone, saying, “You can’t talk people out of their positions on this; they do not want to be confused with facts.” And sure enough, the commenters on his Lingua Franca post proved him right. Perhaps most maddening was this one from someone posting as losemygrip:

Just what the hell is wrong with trying to regularize English and make it a little more consistent? Sounds like a good thing to me. Just because there are inconsistent precedents doesn’t mean we can’t at least try to regularize things. I get so tired of people smugly proclaiming that others are being officious because they want things to make sense.

The desire to fix a problem with the language may seem noble, but in this case the desire stems from a fundamental misunderstanding of the grammar of relative pronouns, and the that/which rule, rather than regularizing the language and making it a little more consistent, actually introduces a rather significant irregularity and inconsistency. The real problem is that few if any grammarians realize that English has two separate systems of relativization: the wh words and that, and they work differently.

If we ignore the various prescriptions about relative pronouns, we find that the wh words (the pronouns who/whom/whose and which, the adverbs where, when, why, whither, and whence, and the where + preposition compounds) form a complete system on their own. The pronouns who and which distinguish between personhood or animacy—people and sometimes animals or other personified things get who, while everything else gets which. But both pronouns function restrictively and nonrestrictively, and so do most of the other wh relatives. (Why occurs almost exclusively as a restrictive relative adverb after reason.)

With all of these relative pronouns and adverbs, restrictiveness is indicated with commas in writing or a small pause in speech. There’s no need for a lexical or morphological distinction to show restrictiveness with who or where or any of the others—intonation or punctuation does it all. There are a few irregularities in the system—for instance, which has no genitive form and must use whose or of which, and who declines for cases while which does not—but on the whole it’s rather orderly.

That, on the other hand, is a system all by itself, and it’s rather restricted in its range. It only forms restrictive relative clauses, and then only in a narrow range of syntactic constructions. It can’t follow a preposition (the book of which I spoke rather than *the book of that I spoke) or the demonstrative that (they want that which they can’t have rather than *they want that that they can’t have), and it usually doesn’t occur after coordinating conjunctions. But it doesn’t make the same personhood distinction that who and which do, and it functions as a relative adverb sometimes. In short, the distribution of that is a subset of the distribution of the wh words. They are simply two different ways to make relative clauses, one of which is more constrained.

Proscribing which in its role as a restrictive relative where it overlaps with that doesn’t make the system more regular—it creates a rather strange hole in the middle of the wh relative paradigm and forces speakers to use a word from a completely different paradigm instead. It actually makes the system irregular. It’s a case of missing the forest for the trees. Grammarians have looked at the distribution of which and that, misunderstood it, and tried to fix it based on their misunderstanding. But if they’d step back and look at the system as a whole, they’d see that the problem is an imagined one. If you think the system doesn’t make sense, the solution isn’t to try to hammer it into something that does make sense; the solution is to figure out what kind of sense it makes. And it makes perfect sense as it is.

I’m sure, as Professor Pullum was, that I’m not going to make a lot of converts. I can practically hear copy editors’ responses: But following the rule doesn’t hurt anything! Some readers will write us angry letters if we don’t follow it! It decreases ambiguity! To the first I say, of course it hurts, in that it has a cost that we blithely ignore: every change a copy editor makes takes time, and that time costs money. Are we adding enough value to the works we edit to recoup that cost? I once saw a proof of a book wherein the proofreader had marked every single restrictive which—and there were four or five per page—to be changed to that. How much time did it take to mark all those whiches for two hundred or more pages? How much more time would it have taken for the typesetter to enter those corrections and then deal with all the reflowed text? I didn’t want to find out the answer—I stetted every last one of those changes. Furthermore, the rule hurts all those who don’t follow it and are therefore judged as being sub-par writers at best or idiots at worst, as Pullum discussed in his Lingua Franca post.

To the second response, I’ve said before that I don’t believe we should give so much power to the cranks. Why should they hold veto power for everyone else’s usage? If their displeasure is such a problem, give me some evidence that we should spend so much time and money pleasing them. Show me that the economic cost of not following the rule in print is greater than the cost of following it. But stop saying that we as a society need to cater to this group and assuming that this ends the discussion.

To the last response: No, it really doesn’t. Commas do all the work of disambiguation, as Stan Carey explains. The car which I drive is no more ambiguous than The man who came to dinner. They’re only ambiguous if you have no faith in the writer’s or editor’s ability to punctuate and thus assume that there should be a comma where there isn’t one. But requiring that in place of which doesn’t really solve this problem, because the same ambiguity exists for every other relative clause that doesn’t use that. Note that Bryan Garner allows either who or that with people; why not allow either which or that with things? Stop and ask yourself how you’re able to understand phrases like The house in which I live or The woman whose hair is brown without using a different word to mark that it’s a restrictive clause. And if the that/which rule really is an aid to understanding, give me some evidence. Show me the results of an eye-tracking study or fMRI or at least a well-designed reading comprehension test geared to show the understanding of relative clauses. But don’t insist on enforcing a language-wide change without some compelling evidence.

The problem with all the justifications for the rule is that they’re post hoc. Someone made a bad analysis of the English system of relative pronouns and proposed a rule to tidy up an imagined problem. Everything since then has been a rationalization to continue to support a flawed rule. Mark Liberman said it well on Language Log yesterday:

This is a canonical case of a self-appointed authority inventing a grammatical theory, observing that elite writers routinely violate the theory, and concluding not that the theory is wrong or incomplete, but that the writers are in error.

Unfortunately, this is often par for the course with prescriptive rules. The rule is taken a priori as correct and authoritative, and all evidence refuting the rule is ignored or waved away so as not to undermine it. Prescriptivism has come a long way in the last century, especially in the last decade or so as corpus tools have made research easy and data more accessible. But there’s still a long way to go.

Update: Mark Liberman has a new post on the that/which rule which includes links to many of the previous Language Log posts on the subject.


Relative What

A few months ago Braden asked in a comment about the history of what as a relative pronoun. (For my previous posts on relative pronouns, see here.) The history of relative pronouns in English is rather complicated, and the system as a whole is still in flux, partly because modern English essentially has two overlapping systems of relativization.

In Old English, there were a few different ways to create a relative pronoun, as this site explains. One way was to use the indeclinable particle þe, another was to use a form of the demonstrative pronoun (roughly equivalent to modern English that/those), and another was to use a demonstrative or personal pronoun followed by þe. Our modern relative that grew out of the use of demonstrative pronouns, though unlike the Old English demonstratives, that does not decline for gender, number, and case.

In the late Old English and Middle English periods, writers and speakers began to use interrogative pronouns as relative pronouns by analogy with French and Latin. It first appeared in texts that were translations from Latin around 1000 AD, but within a couple of centuries it had apparently been naturalized. Other interrogatives became pressed into service as relatives during this time, including who, which, where, when, why, and how. All of these are still in common use in Standard English except for what.

It’s important to note that what is still used as a nominal relative, which means that it does not modify another noun phrase but stands in for a noun phrase and a relative simultaneously, as in We fear what we don’t understand. This could be rephrased as We fear that which we don’t understand or We fear the things that we don’t understand, revealing the nominal and the relative.

But while all the other interrogatives have continued as relatives in Standard English, what as a simple relative pronoun is nonstandard today. Simple relative what is found in the works of Shakespeare and the King James Bible, but at some point in the last three or four centuries it fell out of use in the standard dialect. Unfortunately, I’m not really sure when this happened; the Oxford English Dictionary has citations up through 1740 and then one from 1920 that appears to be dialogue from a novel. Merriam-Webster’s Dictionary of English Usage says that in the US, it’s mainly found in rural areas in the Midland and South. As I told Braden in a response to his comment, I’ve heard it used myself. A couple of months ago I heard a man in church pray for “our leaders what guides and directs us”—not just a beautiful example of relative what, but also an interesting example of nonstandard verb agreement.

As for why simple relative what died out in Standard English, I really have no idea. Jonathan Hope noted that it’s rather unusual of Standard English to allow other interrogatives as relatives but not this one.[1] In some ways, relative what would make more sense than relative which, since what is historically part of the same paradigm as who; what comes from the neuter form of the interrogative or indefinite pronoun in Old English, while who comes from the combined masculine/feminine form, as shown here. And as I said in this post, whose was originally the genitive form for both who and what, so allowing simple relative what would make for a rather tidy paradigm.

Perhaps that’s the problem. Hope and other have argued that standardized languages—or perhaps speakers of standardized languages—tend to resist tidy paradigms. Irregularities creep in and are preserved, and they can be surprisingly resistant to change. Maybe someone reading this has a fuller explanation of just how this particular little wrinkle came to be.

  1. [1] Jonathan Hope, “Rats, Bats, Sparrows and Dogs: Biology, Linguistics and the Nature of Standard English,” in The Development of Standard English, 1300–1800, ed. Laura Wright (Cambridge: University of Cambridge Press, 2000).


What Descriptivism Is and Isn’t

A few weeks ago, the New Yorker published what is nominally a review of Henry Hitchings’ book The Language Wars (which I still have not read but have been meaning to) but which was really more of a thinly veiled attack on what its author, Joan Acocella, sees as the moral and intellectual failings of linguistic descriptivism. In what John McIntyre called “a bad week for Joan Acocella”, the whole mess was addressed multiple times by various bloggers and other writers.* I wanted to write about it at the time but was too busy, but then the New Yorker did me a favor by publishing a follow-up, “Inescapably, You’re Judged by Your Language”, which was equally off-base, so I figured that the door was still open.

I suspected from the first paragraph that Acocella’s article was headed for trouble, and the second paragraph quickly confirmed it. For starters, her brief description of the history and nature of English sounds like it’s based more on folklore than fact. A lot of people lived in Great Britain before the Anglo-Saxons arrived, and their linguistic contributions were effectively nil. But that’s relatively small stuff. The real problem is that she doesn’t really understand what descriptivism is, and she doesn’t understand that she doesn’t understand, so she spends the next five pages tilting at windmills.

Acocella says that descriptivists “felt that all we could legitimately do in discussing language was to say what the current practice was.” This statement is far too narrow, and not only because it completely leaves out historical linguistics. As a linguist, I think it’s odd to describe linguistics as merely saying what the current practice is, since it makes it sound as though all linguists study is usage. Do psycholinguists say what the current practice is when they do eye-tracking studies or other psychological experiments? Do phonologists or syntacticians say what the current practice is when they devise abstract systems of ordered rules to describe the phonological or syntactic system of a language? What about experts in translation or first-language acquisition or computational linguistics? Obviously there’s far more to linguistics than simply saying what the current practice is.

But when it does come to describing usage, we linguists love facts and complexity. We’re less interested in declaring what’s correct or incorrect than we are in uncovering all the nitty-gritty details. It is true, though, that many linguists are at least a little antipathetic to prescriptivism, but not without justification. Because we linguists tend to deal in facts, we take a rather dim view of claims about language that don’t appear to be based in fact, and, by extension, of the people who make those claims. And because many prescriptions make assertions that are based in faulty assumptions or spurious facts, some linguists become skeptical or even hostile to the whole enterprise.

But it’s important to note that this hostility is not actually descriptivism. It’s also, in my experience, not nearly as common as a lot of prescriptivists seem to assume. I think most linguists don’t really care about prescriptivism unless they’re dealing with an officious copyeditor on a manuscript. It’s true that some linguists do spend a fair amount of effort attacking prescriptivism in general, but again, this is not actually descriptivism; it’s simply anti-prescriptivism.

Some other linguists (and some prescriptivists) argue for a more empirical basis for prescriptions, but this isn’t actually descriptivism either. As Language Log’s Mark Liberman argued here, it’s just prescribing on the basis of evidence rather than person taste, intuition, tradition, or peevery.

Of course, all of this is not to say that descriptivists don’t believe in rules, despite what the New Yorker writers think. Even the most anti-prescriptivist linguist still believes in rules, but not necessarily the kind that most people think of. Many of the rules that linguists talk about are rather abstract schematics that bear no resemblance to the rules that prescriptivists talk about. For example, here’s a rather simple one, the rule describing intervocalic alveolar flapping (in a nutshell, the process by which a word like latter comes to sound like ladder) in some dialects of English:

intervocalic alveolar flapping

Rules like these constitute the vast bulk of the language, though they’re largely subconscious and unseen, like a sort of linguistic dark matter. The entire canon of prescriptions (my advisor has identified at least 10,000 distinct prescriptive rules in various handbooks, though only a fraction of these are repeated) seems rather peripheral and inconsequential to most linguists, which is another reason why we get annoyed when prescriptivists insist on their importance or identify standard English with them. Despite what most people think, standard English is not really defined by prescriptive rules, which makes it somewhat disingenuous and ironic for prescriptivists to call us hypocrites for writing in standard English.

If there’s anything disingenuous about linguists’ belief in rules, it’s that we’re not always clear about what kinds of rules we’re talking about. It’s easy to say that we believe in the rules of standard English and good communication and whatnot, but we’re often pretty vague about just what exactly those rules are. But that’s probably a topic for another day.

*A roundup of some of the posts on the recent brouhaha:

Cheap Shot”, “A Bad Week for Joan Acocella”, “Daddy, Are Prescriptivists Real?”, and “Unmourned: The Queen’s English Society” by John McIntyre

Rules and Rules” and “A Half Century of Usage Denialism” by Mark Liberman

Descriptivists as Hypocrites (Again)” by Jan Freeman

Ignorant Blathering at The New Yorker”, by Stephen Dodson, aka Languagehat

Re: The Language Wars” and “False Fronts in the Language Wars” by Steven Pinker

The New Yorker versus the Descriptivist Specter” by Ben Zimmer

Speaking Truth about Power” by Nancy Friedman

Sator Resartus” by Ben Yagoda

I’m sure there are others that I’ve missed. If you know of any more, feel free to make note of them in the comments.


Rules, Evidence, and Grammar

In case you haven’t heard, it’s National Grammar Day, and that seemed as good a time as any to reflect a little on the role of evidence in discussing grammar rules. (Goofy at Bradshaw of the Future apparently had the same idea.) A couple of months ago, Geoffrey Pullum made the argument in this post on Lingua Franca that it’s impossible to talk about what’s right or wrong in language without considering the evidence. Is singular they grammatical and standard? How do you know?

For most people, I think, the answer is pretty simple: you look it up in a source that you trust. If the source says it’s grammatical or correct, it is. If it doesn’t, it isn’t. Singular they is wrong because many authoritative sources say it is. End of story. And if you try to argue that the sources aren’t valid or reliable, you’re labelled an anything-goes type who believes we should just toss all the rules out the window and embrace linguistic anarchy.

The question is, where did these sources get their authority to say what’s right and wrong?

That is, when someone says that you should never use they as a singular pronoun or start a sentence with hopefully or use less with count nouns, why do you suppose that the rules they put forth are valid? The rules obviously haven’t been inscribed on stone tablets by the finger of the Lord, but they have to come from somewhere. Every language is different, and languages and constantly changing, so I think we have to recognize that there is no universal, objective truth when it comes to grammar and usage.

David Foster Wallace apparently fell into the trap of thinking that there was, unfortunately. In his famous Harper’s article “Tense Present: Democracy, English, and the Wars over Usage,” he quotes the introduction to The American College Dictionary, which says, “A dictionary can be an “authority” only in the sense in which a book of chemistry or of physics or of botany can be an “authority”: by the accuracy and the completeness of its record of the observed facts of the field examined, in accord with the latest principles and techniques of the particular science.”

He retorts,

This is so stupid it practically drools. An “authoritative” physics text presents the results of physicists’ observations and physicists’ theories about those observations. If a physics textbook operated on Descriptivist principles, the fact that some Americans believe that electricity flows better downhill (based on the observed fact that power lines tend to run high above the homes they serve) would require the Electricity Flows Better Downhill Theory to be included as a “valid” theory in the textbook—just as, for Dr. Fries, if some Americans use infer for imply, the use becomes an ipso facto “valid” part of the language.

The irony of his first sentence is almost overwhelming. Physics is a set of universal laws that can be observed and tested, and electricity works regardless of what anyone believes. Language, on the other hand, is quite different. In fact, Wallace tacitly acknowledges the difference—without explaining his apparent contradiction—immediately after: “It isn’t scientific phenomena they’re tabulating but rather a set of human behaviors, and a lot of human behaviors are—to be blunt—moronic. Try, for instance, to imagine an ‘authoritative’ ethics textbook whose principles were based on what most people actually do.”[1]

Now here he hits on an interesting question. Any argument about right or wrong in language ultimately comes down to one of two options: it’s wrong because it’s absolutely, objectively wrong, or it’s wrong because arbitrary societal convention says it’s wrong. The former is untenable, but the latter doesn’t give us any straightforward answers. If there is no objective truth in usage, then how do we know what’s right and wrong?

Wallace tries to make the argument about ethics; sloppy language leads to real problems like people accidentally eating poison mushrooms. But look at his gargantuan list of peeves and shibboleths on the first page of the article. How many of them lead to real ethical problems? Does singular they pose any kind of ethical problem? What about sentential hopefully or less with count nouns? I don’t think so.

So if there’s no ethical problem with disputed usage, then we’re still left with the question, what makes it wrong? Here we get back to Pullum’s attempt to answer the question: let’s look at the evidence. And, because we can admit, like Wallace, that some people’s behavior is moronic, let’s limit ourselves to looking at the evidence from those speakers and writers whose language can be said to be most standard. What we find even then is that a lot of the usage and grammar rules that have been put forth, from Bishop Robert Lowth to Strunk and White to Bryan Garner, don’t jibe with actual usage.

Edward Finegan seizes on this discrepancy in an article a few years back. In discussing sentential hopefully, he quotes Garner as saying that it is “all but ubiquitous—even in legal print. Even so, the word received so much negative attention in the 1970s and 1980s that many writers have blacklisted it, so using it at all today is a precarious venture. Indeed, careful writers and speakers avoid the word even in its traditional sense, for they’re likely to be misunderstood if they use it in the old sense”[2] Finegan says, “I could not help but wonder how a reflective and careful analyst could concede that hopefully is all but ubiquitous in legal print and claim in the same breath that careful writers and speakers avoid using it.”[3]

The problem when you start questioning the received wisdom on grammar and usage is that you make a lot of people very angry. In a recent conversation on Twitter, Mignon Fogarty, aka Grammar Girl, said, “You would not believe (or maybe you would) how much grief I’m getting for saying ‘data’ can sometimes be singular.” I responded, “Sadly, I can. For some people, grammar is more about cherished beliefs than facts, and they don’t like having them challenged.” They don’t want to hear arguments about authority and evidence and deriving rules from what educated speakers actually use. They want to believe that there’s some deeper truths that justify their preferences and peeves, and that’s probably not going to change anytime soon. But for now, I’ll keep trying.

  1. [1] David Foster Wallace, “Tense Present: Democracy, English, and the Wars over Usage,” Harper’s Monthly, April 2001, 47.
  2. [2] “Bryan A. Garner, A Dictionary of Modern Legal Usage, 2nd ed. (New York: Oxford University Press, 1995).
  3. [3] Edward Finegan, “Linguistic Prescription: Familiar Practices and New Perspectives,” Annual Review of Applied Linguistics (2003) 23, 216.


More on That

As I said in my last post, I don’t think the distribution of that and which is adequately explained by the restrictive/nonrestrictive distinction. It’s true that nearly all thats are restrictive (with a few rare exceptions), but it’s not true that all restrictive relative pronouns are thats and that all whiches are nonrestrictive, even when you follow the traditional rule. In some cases that is strictly forbidden, and in other cases it is disfavored to varying degrees. Something that linguistics has taught me is that when your rule is riddled with exceptions and wrinkles, it’s usually sign that you’ve missed something important in your analysis.

In researching the topic for this post, I’ve learned a couple of things: (1) I don’t know syntax as well as I should, and (2) the behavior of relatives in English, particularly that, is far more complex than most editors or pop grammarians realize. First of all, there’s apparently been a century-long argument over whether that is even a relative pronoun or actually some sort of relativizing conjunction or particle. (Some linguists seem to prefer the latter, but I won’t wade too deep into that debate.) Previous studies have looked at multiple factors to explain the variation in relativizers, including the animacy of the referent, the distance between the pronoun and its referent, the semantic role of the relative clause, and the syntactic role of the referent.

It’s often noted that that can’t follow a preposition and that it doesn’t have a genitive form of its own (it must use either whose or of which), but no usage guide I’ve seen ever makes mention of the fact that this pattern follows the accessibility hierarchy. That is, in a cross-linguistic analysis, linguists have found an order to the way in which relative clauses are formed. Some languages can only relativize subjects, others can do subjects and verbal objects, yet others can do subjects, verbal objects, and oblique objects (like the objects of prepositions), and so on. For any allowable position on the hierarchy, all positions to the left are also allowable. The hierarchy goes something like this:

subject ≥ direct object ≥ indirect object ≥ object of stranded preposition ≥ object of fronted preposition ≥ possessor noun phrase ≥ object of comparative particle

What is interesting is that that and the wh- relatives, who and which, occupy overlapping but different portions of the hierarchy. Who and which can relativize anything from subjects to possessors and possibly objects of comparative particles, though whose as the genitive form of which seems a little odd to some, and both sound odd if not outright ungrammatical with comparatives, as in The man than who I’m taller. But that can’t relativize objects of fronted prepositions or anything further down the scale.

Strangely, though, there are things that that can do that who and which can’t. That can sometimes function as a sort of relative adverb, equivalent to the relative adverbs why, where, or when or to which with a preposition. That is, you can say The day that we met, The day when we met, or The day on which we met, but not The day which we met. And which can relativize whole clauses (though some sticklers consider this ungrammatical), while that cannot, as in This author uses restrictive “which,” which bothers me a lot.

So what explains the differences between that and which or who? Well, as I mentioned above, some linguists consider that not a pronoun but a complementizer or conjunction (perhaps a highly pronominal one), making it more akin to the complementizer that, as in He said that relativizers were confusing. And some linguists have also proposed different syntactic structures for restrictive and nonrestrictive clauses, which could account for the limitation of that to restrictive clauses. If that is not a true pronoun but a complementizer, then that could account for its strange distribution. It can’t appear in nonrestrictive clauses, because they require a full pronoun like which or who, and it can’t appear after prepositions, because those constructions similarly require a pronoun. But it can function as a relative adverb, which a regular relative pronoun can’t do.

As I argued in my previous post, it seems that which and that do not occupy separate parts of a single paradigm but are part of two different paradigms that overlap. The differences between them can be characterized in a few different ways, but for some reason, grammarians have seized on the restrictive/nonrestrictive distinction and have written off the rest as idiosyncratic exceptions to the rule or as common errors (when they’ve addressed those points at all).

The proposal to disallow which in restrictive relative clauses, except in the cases where that is ungrammatical—sometimes called Fowler’s rule, though that’s not entirely accurate—is based on the rather trivial observation that all thats are restrictive and that all nonrestrictives are which. It then assumes that the converse is true (or should be) and tries to force all restrictives to be that and all whiches to be nonrestrictive (except for all those pesky exceptions, of course).

Garner calls Fowler’s rule “nothing short of brilliant,”[1] but I must disagree. It’s based on a rather facile analysis followed by some terrible logical leaps. And insisting on following a rule based on bad linguistic analysis is not only not helpful to the reader, it’s a waste of editors’ time. As my last post shows, editors have obviously worked very hard to put the rule into practice, but this is not evidence of its utility, let alone its brilliance. But a linguistic analysis that could account for all of the various differences between the two systems of relativization in English? Now that just might be brilliant.


Herbert F. W. Stahlke, “Which That,” Language 52, no. 3 (Sept. 1976): 584–610
Johan Van Der Auwera, “Relative That: A Centennial Dispute,” Journal of Lingusitics 21, no. 1 (March 1985): 149–79
Gregory R. Guy and Robert Bayley, “On the Choice of Relative Pronouns in English,” American Speech 70, no. 2 (Summer 1995): 148–62
Nigel Fabb, “The Difference between English Restrictive and Nonrestrictive Relative Clauses,” Journal of Linguistics 26, no. 1 (March 1990): 57–77
Robert D. Borsley, “More on the Difference between English Restrictive and Nonrestrictive Relative Clauses,” Journal of Linguistics 28, no. 1 (March 1992), 139–48

  1. [1] Garner’s Modern American Usage, 3rd ed., s.v. “that. A. And which.”


Rules, Regularity, and Relative Pronouns

The other day I was thinking about relative pronouns and how they get so much attention from usage commentators, and I decided I should write a post about them. I was beaten to the punch by Stan Carey, but that’s okay, because I think I’m going to take it in a somewhat different direction. (And anyway, great minds think alike, right? But maybe you should read his post first, along with my previous post on who and that, if you haven’t already.)

I’m not just talking about that and which but also who, whom, and whose, which is technically a relative possessive adjective. Judging by how often relative pronouns are talked about, you’d assume that most English speakers can’t get them right, even though they’re among the most common words in the language. In fact, in my own research for my thesis, I’ve found that they’re among the most frequent corrections made by copy editors.

So what gives? Why are they so hard for English speakers to get right? The distinctions are pretty clear-cut and can be found in a great many usage and writing handbooks. Some commentators even judgementally declare, “There’s a useful distinction here, and it’s lazy or perverse to pretend otherwise.” But is it really useful, and is it really lazy and perverse to disagree? Or is it perverse to try to inflict a bunch of arbitrary distinctions on speakers and writers?

And arbitrary they are. Many commentators act as if the proposed distinctions between all these words would make things tidier and more regular, but in fact it makes the whole system much more complicated. On the one hand, we have the restrictive/nonrestrictive distinction between that and which. On the other hand, we have the animate/inanimate (or human/nonhuman, if you want to be really strict) distinction between who and that/which. And on the other other hand, there’s the subject/object distinction between who and whom. But there’s no subject/object distinction with that or which, except when it’s the object of a preposition—then you have to use which, unless the preposition is stranded, in which case you can use that. And on the final hand, some people have proscribed whose as an inanimate or nonhuman relative possessive adjective, recommending constructions with of which instead, though this rule isn’t as popular, or at least not as frequently talked about, as the others. (How many hands is that? I’ve lost count.)

Simple, right? To make it all a little clear, I’ve even put it into a nice little table.

The proposed relative pronoun system

This is, in a nutshell, a very lopsided and unusual system. In a comment on my who/that post, Elaine Chaika says, “No natural grammar rule would work that way. Ever.” I’m not entirely convinced of that, because languages can be surprising in the unusual distinctions they make, but I agree that it is at the least typologically unusual.

“But we have to have rules!” you say. “If we don’t, we’ll have confusion!” But we do have rules—just not the ones that are proposed and promoted. The system we really have, in absence of the prescriptions, is basically a distinction between animate who and inanimate which with that overlaying the two. Which doesn’t make distinctions by case, but who(m) does, though this distinction is moribund and has probably only been kept alive by the efforts of schoolteachers and editors.

Whom is still pretty much required when it immediately follows a preposition, but not when the preposition is stranded. Since preposition stranding is extremely common in speech and increasingly common in writing, we’re seeing less and less of whom in this position. Whose is still a little iffy with inanimate referents, as in The house whose roof blew off, but many people say this is alright. Others prefer of which, though this can be awkward: The house the roof of which blew off.

That is either animate or inanimate—only who/which make that distinction—and can be either subject or object but cannot follow a preposition or function as a possessive adjective or nonrestrictively. If the preposition is stranded, as in The man that I gave the apple to, then it’s still allowed. But there’s no possessive thats, so you have to use whose of of which. Again, it’s clearer in table form:

The natural system of relative pronouns

The linguist Jonathan Hope wrote that several distinguishing features of Standard English give it “a typologically unusual structure, while non-standard English dialects follow the path of linguistic naturalness.” He then muses on the reason for this:

One explanation for this might be that as speakers make the choices that will result in standardisation, they unconsciously tend towards more complex structures, because of their sense of the prestige and difference of formal written language. Standard English would then become a ‘deliberately’ difficult language, constructed, albeit unconsciously, from elements that go against linguistic naturalness, and which would not survive in a ‘natural’ linguistic environment.[1]

It’s always tricky territory when you speculate on people’s unconscious motivations, but I think he’s on to something. Note that while the prescriptions make for a very asymmetrical system, the system that people naturally use is moving towards a very tidy and symmetrical distribution, though there are still a couple of wrinkles that are being worked out.

But the important point is that people already follow rules—just not the ones that some prescriptivists think they should.

  1. [1] “Rats, Bats, Sparrows and Dogs: Biology, Linguistics and the Nature of Standard English,” in The Development of Standard English, 1300–1800, ed. Laura Wright (Cambridge: University of Cambridge Press, 2000), 53.